Open pravi opened 4 months ago
Today talk-staging.puri.sm also failed. For now used a work around - replacing tls with an SSH tunnel. SSH tunnel is created from synapse server to ldap server on 389 and in synapse configuration changed connection to localhost:389 and start_tls: false. This is working, new sessions are working.
https://github.com/nodejs/node/issues/27825#issuecomment-495145517 seems to be the issue we have here as well.
Description
ldap authentication abruptly stopped working. This was working for many years and the breakage seems to be after updating to 1.108.0. Strangely same version of synapse on talk-staging.puri.sm can authenticate against the same ldap server. Other services are authenticating correctly with the same ldap server.
I could reproduce the error when cloning the server and with a new blank database and changing server name to talk-troubles.puri.sm in nginx and homeserver.yaml
I tried to clone talk-staging.puri.sm and replace database and server name with talk.puri.sm, but the error is still there.
Steps to reproduce
Homeserver
talk.puri.sm
Synapse Version
1.108.0
Installation Method
pip (from PyPI)
Database
PostgreSQL
Workers
Multiple workers
Platform
Digital Ocean cloud VM with Debian GNU/Linux bookworm/stable.
Configuration
using ldap authentication
Relevant log output
Anything else that would be useful to know?
This was reproduced on 3 servers but one old server was working with same synapse version and ldap server.